This guide covers each stage of the FRI process: what to do, in what order, and what to record, so your team produces a clear, defensible result every time.
What is a Final Random Inspection?
A Final Random Inspection (FRI) is a structured quality check performed on randomly selected units from a finished production lot. The goal is to verify that what is in those cartons matches the buyer's specifications (workmanship, labeling, packaging, functional performance, and carton markings) before shipment is approved.
The 'random' part is not optional. Units are selected using a sampling plan (typically based on the AQL standard) so no one on the factory floor can hand-pick the best pieces for inspection.
When to run the FRI
Run the FRI after 100% of the order is finished and packed into cartons. The lot must be ready to ship, not half-packed, not still moving through the line. Inspecting early means you may miss defects introduced during final packing.
A common exception: when a buyer requires an inline inspection at a set completion threshold. In that case, run inline at 80% complete and FRI at 100%. Both inspections produce separate reports.
Step 1: Set the AQL threshold before the inspection starts
Before the inspector touches a single unit, the acceptance criteria must be agreed and written down. The AQL (Acceptable Quality Limit) defines how many defects in the sample will cause the lot to fail.
- Critical defects: AQL 0, one critical defect and the lot fails
- Major defects: AQL 2.5, workmanship or functional issues the buyer will reject
- Minor defects: AQL 4.0, cosmetic issues that do not affect function
These numbers must be agreed before the inspection begins. Adjusting the threshold after reviewing the sample is how disputes start.
Step 2: Calculate the required sample size
Sample size comes from the AQL table, based on the lot size and the inspection level. For most consumer goods exports, General Inspection Level II is the standard starting point.
A lot of 3,201–10,000 units at Level II produces a required sample size of 200 units. Pull the number from the table, record it on the inspection sheet, and do not adjust it because the lot looks clean at first glance.
- Lot size 2–8 units → sample 2
- Lot size 151–280 units → sample 32
- Lot size 1,201–3,200 units → sample 125
- Lot size 3,201–10,000 units → sample 200
- Lot size 10,001–35,000 units → sample 315
Step 3: Select units at random
Random selection means the inspector pulls units from different positions in the lot (top cartons, bottom cartons, middle of the stack) without anyone on the factory side guiding the selection.
If the lot is palletized, pull from at least three different pallet positions. If cartons are numbered, use a random number generator to choose which cartons to open. The sample must represent the full lot, not the visible or most accessible section.
Step 4: Inspect each unit against the specification
Each selected unit gets checked against the agreed spec sheet and the approved production sample. The inspector works from a structured checklist, not a blank notebook.
- Workmanship: seams, stitching, surface finish, assembly quality, hardware attachment
- Functional tests: zippers, closures, hinges, electrical function, whatever the product requires
- Appearance: color accuracy, print registration, logo placement, label content and language
- Packaging: inner box, barcode scan, carton labeling, quantity per carton
- Measurements: key dimensions checked against the specification sheet
Any deviation from spec gets logged immediately, not at the end of the walk.
Step 5: Record every defect with a photo
Every defect found gets logged at the moment of finding. The inspector records the unit identifier, the defect description, and the severity classification (critical, major, or minor).
Photos are mandatory for every defect. A written note without a photo is not proof: it is an opinion. The photo locks the finding to a specific unit at a specific point in time and removes ambiguity when the report reaches the buyer or the factory QC team.
Do not wait until the end of the inspection to document. Memory fades. Photo quality degrades. Log it as you find it.
Step 6: Compare totals against the AQL limits and issue the decision
Once the full sample is checked, total the defect counts by severity and compare them against the AQL table limits.
Example: a sample of 200 units at AQL 2.5 for major defects allows a maximum of 10 major defects before the lot fails. Finding 11 major defects means the lot fails, regardless of how the other 189 units looked.
The decision is one of three outcomes:
- Pass: defect counts are within the agreed AQL limits. The lot can proceed to shipment.
- Hold: defects were found and the factory has agreed to sort a portion of the lot. Re-inspection is required before shipment is released.
- Fail: defect counts exceed the AQL limits. The lot does not ship. Root cause investigation and corrective action must happen before re-inspection is scheduled.
State the decision clearly on the report. The decision is not a negotiation: it follows directly from the numbers.
Step 7: Share the report before the goods move
The complete inspection report (defect log, photos, measurement results, sample size, and the pass/fail decision) must reach the factory team and the buyer before the shipment is released.
Sending the report after the goods have already shipped turns it into a formality. The report's value is in the decision it supports. If it does not arrive in time to stop a bad shipment, it has not done its job.
How to make FRIs consistent across your team
The biggest risk in a manual FRI process is variation between inspectors. One inspector logs every minor defect. Another ignores cosmetic issues. The result is that the same lot produces different outcomes depending on who ran the walk.
Lock consistency in through three practices:
- Use the same digital checklist for every FRI, not a blank form each inspector fills in their own way
- Require photos for every defect finding, so there is no room for subjective interpretation
- Record the AQL threshold and sample size on every report, so the basis for the decision is always documented
When every inspector follows the same process and records the same evidence, the FRI result means the same thing regardless of who ran it.
Key takeaways
- Set the AQL threshold in writing before the inspection begins, never after reviewing the sample.
- Pull units from random positions in the lot: top, bottom, and middle of the stack.
- Log every defect immediately with a photo. Notes written after the walk are not proof.
- The pass/fail decision is a calculation against agreed limits, not a judgment call.
- Share the complete report with the factory and the buyer before the goods ship.
Put this into practice with KaizenQ
KaizenQ gives your team a structured digital inspection workflow — defects logged with photos on the floor, a live report ready to share the moment the inspection ends. No manual compilation. No end-of-day rework.